System and method for mapping activation waves to guide treatment of biological rhythm disorders

Abstract

A heart treatment system is disclosed capable of guiding a device towards one or more critical regions of interest by sensing signals from tissue. If a critical region is not present at the current location of sensed signals, the system is capable of indicating a guidance direction in which to navigate to reach one or more critical regions. When stopping rules for direction are met, treatment can be applied to said region of interest by thermal or non-thermal energy delivery. Signals are again sensed and analyzed to assess the impact of treatment. This process is repeated until all critical regions of interest are treated. In some embodiments, all functionality is provided by a single sensing and treating device coupled with a display device and analytical software.

What is claimed is: 
     
       1. A method for guiding therapy for a heart rhythm disorder comprising:
 receiving a set of electrical signals of a human heart measured by a plurality of sensing electrodes on a catheter in contact with the human heart, wherein each sensing electrode of the plurality of sensing electrodes generates one electrical signal, and wherein the sensing electrodes are disposed in a known spatial configuration; 
 applying an activation detection model to identify one or more activations in each electrical signal that represent activation times for the heart rhythm disorder, wherein the activation detection model is a deep-learning neural network trained using a training set of electrical signals measured from one or more human hearts from one or more training subjects, the training set of electrical signals annotated with ground truth activations, wherein applying the activation detection model to each electrical signal of the set of electrical signals comprises:
 applying the activation detection model trained as the deep-learning neural
 network to convert each electrical signal of the set of electrical signals into an activation 
 likelihood timeseries, and 
 
 identifying one or more peaks in the activation likelihood timeseries above a threshold activation likelihood as the one or more activations in the electrical signal; 
 
 determining one or more beats of the heart rhythm disorder, wherein each
 beat of the one or more beats is determined by:
 determining a confidence that pairwise activations from neighboring sensing electrodes in the spatial configuration belong to one beat based on the spatial configuration of the sensing electrodes and tissue conduction velocity, 
 generating a graph comprising the pairwise activations and the confidences as nodes of the graph, and 
 clustering a subset of the pairwise activations across the set of electrical signals as one
 beat by linking nodes of the graph with confidences above a threshold; 
 
 
 
 calculating a predominant direction of the heart rhythm disorder based on the one or more beats; 
 determining a guidance direction based on the predominant direction of the heart rhythm disorder, wherein the guidance direction directs the catheter towards a critical region of the heart rhythm disorder; 
 displaying a graphical user interface depicting a visual representation of the catheter and the guidance direction on an electronic display of a computing device operated by a healthcare provider; and 
 transmitting control instructions to the catheter, wherein the control instructions cause the catheter to deliver ablation energy at the critical region of the heart rhythm disorder. 
 
     
     
       2. The method of  claim 1 , wherein the activation detection model is trained by:
 identifying the threshold activation likelihood by adapting to the heart rhythm disorder, based on a fixed blanking period, or based on an external signal source. 
 
     
     
       3. The method of  claim 2 , further comprising:
 determining, for each electrical signal of the set of electrical signals, a signal quality score based on a strength of
 activations for each beat of the one or more beats identified in the electrical signal. 
 
 
     
     
       4. The method of  claim 3 , further comprising:
 labeling a first electrical signal as unusable based on the signal quality score for the first electrical signal being below a threshold score. 
 
     
     
       5. The method of  claim 4 , further comprising, in response to labeling the first electrical signal as unusable, reconstructing a synthetic electrical signal for use in place of the first electrical signal based on electrical signals of neighboring sensing electrodes. 
     
     
       6. The method of  claim 1 , wherein the activation detection model is further trained using a second training set of action potential recordings annotated with activations. 
     
     
       7. The method of  claim 1 , wherein determining the one or more beats of the heart rhythm disorder comprises defining, for each pair of neighboring sensing electrodes, a time between consecutive activations based on a spatial distance between the neighboring sensing electrodes and tissue conduction velocity. 
     
     
       8. The method of  claim 7 , wherein the spatial distance between the neighboring sensing electrodes is determined by geometric estimation from the tissue conduction velocity into a symmetrical representation centered around each activation time of the activation times. 
     
     
       9. The method of  claim 1 , wherein the activation detection model is configured to identify the one or more activations in each electrical signal of the set of electrical signals based on interpolation or extrapolation from identified activations in the electrical signals at neighboring sensing electrodes. 
     
     
       10. The method of  claim 1 , wherein determining each beat of the one or more beats comprises identifying a sequential ordering of the set of electrical signals by calculating a gradient based on time shift that maximizes cross-correlation between the electrical signals. 
     
     
       11. The method of  claim 1 , wherein determining the predominant direction comprises:
 determining a wave direction for each beat of the one or more beats; and 
 integrating wave directions of the one or more beats. 
 
     
     
       12. The method of  claim 11 , wherein the predominant direction is a spatial average of wave directions representing distinct activations. 
     
     
       13. The method of  claim 11 , wherein determining the predominant direction further comprises:
 identifying an aggregated temporal difference of activation times at each electrode of the plurality of sensing electrodes for
 successive waves; and 
 
 determining the predominant direction from the aggregated temporal differences at the electrodes. 
 
     
     
       14. The method of  claim 1 , further comprising determining that the catheter is positioned at the critical region of the heart rhythm disorder based on the predominant direction. 
     
     
       15. The method of  claim 14 , wherein determining that the catheter is positioned at the critical region is based on determining that predominant directions at successive time periods converge to one location. 
     
     
       16. The method of  claim 15 , wherein determining that the catheter is positioned at the critical region is based on determining that directions of sub-regions of the catheter indicate a critical site. 
     
     
       17. The method of  claim 14 , wherein determining that the catheter is positioned at the critical region is based on determining that a critical feature for the heart rhythm disorder is detected from the electrical signals. 
     
     
       18. The method of  claim 14 , wherein determining that the catheter is positioned at the critical region is based on determining that the predominant direction of the heart rhythm disorder is a near-zero vector. 
     
     
       19. The method of  claim 1 , further comprising:
 terminating operation of the catheter in response to determining that the set of electrical signals indicate activation likelihood below a threshold. 
 
     
     
       20. The method of  claim 1 , wherein the plurality of electrodes is arranged in an array with known spacing in a rectilinear configuration, a radially emanating configuration, a spherical configuration, or a spiral configuration. 
     
     
       21. The method of  claim 1 , wherein the predominant direction covers a sub-region of the catheter inclusive of a subset of the sensing electrodes. 
     
     
       22. The method of  claim 21 , wherein the subset of the sensing electrodes correspond to a subset of activations in a defined period of time. 
     
     
       23. The method of  claim 1 , further comprising:
 determining a confidence score for the predominant direction based on the one or more beats. 
 
     
     
       24. The method of  claim 23 , wherein the confidence score is based on directional variance, temporal variance, spatial variance, method variance, wave variance, signal quality, or some combination thereof. 
     
     
       25. A method for guiding therapy for a heart rhythm disorder comprising:
 receiving a set of electrical signals of a human heart measured by a plurality of sensing electrodes on a catheter in contact with the human heart, wherein each sensing electrode of the plurality of sensing electrodes generates one electrical signal, and wherein the sensing electrodes are disposed in a two-dimensional array; 
 applying an activation detection model to identify one or more activations in each electrical signal that represent activation times for the heart rhythm disorder, wherein the activation detection model is a deep-learning neural network trained using a training set of electrical signals measured from one or more human hearts from one or more training subjects, the training set of electrical signals annotated with ground truth activations wherein applying the activation detection model to each electrical signal of the set of electrical signals comprises:
 applying the activation detection model trained as the deep-learning neural
 network to convert each electrical signal of the set of electrical signals into an activation 
 likelihood timeseries, and 
 
 identifying one or more peaks in the activation likelihood timeseries above a threshold activation likelihood as the one or more activations in the electrical signal; 
 
 determining one or more beats of the heart rhythm disorder, wherein each
 beat of the one or more beats is determined by:
 determining a confidence that pairwise activations from neighboring sensing electrodes in the two-dimensional array belong to one beat based on a spatial configuration of the sensing electrodes and tissue conduction velocity, 
 generating a graph comprising the pairwise activations and the confidences as nodes of the graph, and 
 clustering a subset of the pairwise activations across the set of electrical signals as one
 beat by linking nodes of the graph with confidences above a threshold; 
 
 
 
 calculating a predominant direction of the heart rhythm disorder based on the one or more beats; 
 displaying a graphical user interface depicting a visual representation of the catheter and the predominant direction on an electronic display of a computing device operated by a healthcare provider; and 
 transmitting control instructions to the catheter, wherein the control instructions cause the catheter to deliver ablation energy at a critical region of the heart rhythm disorder.